1. Openness & Flexibility (self‑hosting, provider variance, open weights)
- “Is the model you trained available as open weights?” — redrove
- “Can it route to locally or LAN hosted Qwen or some other open weights model?” — gitowiec
- “How do you handle provider variance on OpenRouter for the opensource models? Or do you use your own hosted version to mitigate this?” — ajspig1
- “We plug into any harness (e.g. Claude Code, Codex, OpenCode, Pi).” — adchurch
2. Cost‑Effectiveness / Budget & Efficiency vs. Frontier Models
- “Does this allow for a predefined budget?” — 1minusp
- “If you are training on data labeled by frontier models, how do you expect to exceed the performance of frontier models, other than in the cost dimension by recognizing simpler problems and routing to cheaper models?” — svnt
- “From my test, 6.1 Sol is a lot more token efficient than 6 Sol while being similar to Astra in performance…” — YuechenLi
- “We aren't incentivized to route to our own model, we're incentivized to route to the best model whatever it may be.” — adchurch
3. Similarity to Existing AI Coding Assistants (Cursor auto mode, Copilot)
- “How would you say this compares to Cursor's auto mode?” — thefourthchime
- “And similarly, Copilot’s Auto mode?” — aschla
- “Absolutely! Conceptually very similar to Cursor's auto mode.” — adchurch
- “How do you define the model buckets, and what happens when a session genuinely needs a model that isn't in the bucket the HMM picked?” — jamesforestwest